Efficient simulation of automotive multi-physical systems by optimization with evolutionary algorithms

Microelectronic products for automotive applications are characterized by an ever-increasing level of integration complexity concerning electronic and multi-physical subsystems. Since the functional verification of a developed system is normally done by transient simulations, system engineers have to integrate models of different physical disciplines efficiently into one simulation environment. This paper discusses the parameter optimization of a simulation protocol for a multi-simulator co-simulation toot called InSiMS, using a multiple-objective evolutionary algorithm. Two objective functions are used to minimize the error for the transient signals and to maximize the simulation speedup compared to a set of nominal simulation runs. The benefit of the proposed method will be illustrated by the simulation of a fuel injection system, consisting of a hydromechanic injector and an electromagnetic controller.